1022 Background: Randomized trials show that non-steroidal anti-inflammatory drugs (NSAIDs) reduce the risk of pre-cancerous polyps and may reduce the risk of invasive colorectal cancer (CRC), but studies of effectiveness have not been reported. Given the ubiquity of NSAIDS, existing population-level data sources may be leveraged to provide a window into the effectiveness of NSAIDs in CRC prevention. Methods: With data from the National Center for Health Statistics’ National Ambulatory Medical Care Survey, we established the suitability of the diagnosis of osteoarthritis (OA) as an instrument or marker of NSAID therapy among the elderly. With data from the National Cancer Institute’s SEER-Medicare program, we then identified elderly Medicare beneficiaries diagnosed with CRC in 1995 and compared them to elderly Medicare beneficiaries without histories of CRC according to antecedent OA diagnoses in Medicare claims files. Using logistic regression, we estimated the odds of CRC diagnosis among elderly Medicare beneficiaries with histories of OA compared to those without. As a control analysis, we also estimated the odds of a breast cancer diagnosis among those elderly female Medicare beneficiaries with histories of OA compared to those without. Results: Within the SEER-Medicare data files, we identified 4,599 elderly patients with a SEER diagnosis of invasive CRC and 100,270 patients without a SEER diagnosis of invasive CRC. Individuals with a diagnosis of OA in Medicare claims in the prior three years had an 18% lower risk of being diagnosed with CRC than individuals whose claims did not reflect antecedent OA (OR 0.82, 95% CI: 0.77–0.88). No association between antecedent OA and subsequent breast cancer (OR 0.99, 95% CI: 0.92–1.05) was found. Conclusions: In this case-control study that treats certain ICD-9 codes within Medicare claims as instruments of NSAID use, we find that elderly Medicare beneficiaries with ICD-9 codes for OA have an18% lower odds of being diagnosed with invasive CRC. Further, the magnitude of the CRC odds reduction is similar to what would be expected from the excess NSAID use alone. No significant financial relationships to disclose.